No objections having been raised, I've submitted a patch (mousein-non-exclusive) to remove this restriction.
Next, the same question about /dev/kbin - does that need to be exclusive open? I think not. Slightly more complicated, because kbin can receive a pair of inputs (escape + scancode) and pass them one at a time to the kbd driver. But that is already a potential bug, not prevented by exclusive open - usbd can start driver processes for two usb keyboards, multiplexing their inputs through one fd to kbin, and on a multiprocessor the escapes might be incorrectly interleaved. My proposal is to remove the exclusive open, and instead put a qlock around the loop which sends codes from kbin to kbd.
